ABSTRACT
Aims: To study the diversity of traditional rice genotypes in Sri Lanka using cluster analysis and principle component analysis. Study Design: The experiment was carried out using one hundred rice genotypes with six modern rice cultivars and ninety four traditional rice cultivars. Rice genotypes were planted according to a randomized complete block design with four replications and 20 plants per plot with 15 cm X 20 cm spacing. Place and Duration of Study: A field experiment was carried out during 2011/2012 Maha season and 2012 Yala season at Faculty of Agriculture, University of Ruhuna, Sri Lanka. Methodology: Plant height (cm), number of tillers/plant and number of productive tillers/plant were measured before harvesting. Panicle length (cm), panicle weight (g), number of spikelets/panicle, number of fertile spikelets/panicle, 100 grain weight (g) and yield/plant (g) were measured after harvesting and drying of grains for 14% moisture content. Principal component analysis (PCA) and cluster analysis were performed using SPSS statistical software. Results: Among nine studied variables three principal components exhibited more than one Eigen value and showed about 89.6 % variability. The principal components (PC) 1, 2 and 3 had 51.07%, 22.08% and 16.46% variability among the genotypes for the evaluated traits respectively. The first PC was more related to panicle weight, number of spikelets/panicle, number of fertile spikelets/panicle, spikelet fertility percentage and yield (g/plant). Number of tillers/plant, number of productive tillers/plant and yield (g/plant) were more related traits in the second principal component. The highest contribution in third principal component was from the panicle weight, 100 grain weight and yield (g/plant). Based on the nine yield and yield attributing characters, the genotypes were grouped in to seven clusters in cluster analysis. The genotypes under cluster V recorded the highest divergence among them as it exhibited the highest intra-cluster distance. The lowest intracluster distance was recorded in the cluster VI. The modern rice cultivar BG 379/2 was fallen in to the cluster VI with 3 traditional rice cultivars namely Karayal I, Bathkiri el and Hondarawala. Conclusion: One hundred rice genotypes were grouped in to divergent groups by principle component analysis and cluster analysis. This clustering pattern can be used for the selection of parental materials with diverse characters.